To address the need for precise perception of rapidly maneuvering and diverse unmanned aerial vehicle(UAV) models in complex airspace environments, a UAV signal recognition method based on time-frequency multi-channel features is proposed. In the signal processing stage, the UAV signals undergo detrending and denoising to eliminate direct current components and low-frequency drift, thereby reducing frequency-domain interference. The discrete Fourier transform is then applied to extract time-frequency information, followed by spectrum splicing and amplitude normalization to construct energy spectrum data. To address spatial heterogeneity, multi-channel processing is performed along the frequency domain. For the model architecture, a multi-feature channel signal tensorization strategy is proposed based on an MC-2DCNN, precisely aligning with the model’s multi-channel convolutional characteristics. Through spatial decoupling and structured reorganization of frequency-domain features, hierarchical extraction and differentiated modeling of cross-band features are achieved while preserving feature topological adjacency, leveraging channel separation mechanisms to enhance model robustness and generalization. Experiments demonstrate that the proposed method achieves recognition accuracies of 98.15% and 96.32% for UAV models and flight modes, respectively, within the -10 dB to 10 dB noise range. Under aliasing noise scenarios involving electromagnetic noise, comb spectrum, and blocking interference, it attains accuracies of 91.97% and 90.56%, meeting intelligent recognition requirements for UAV detection systems in both civilian and military applications.
YANXingwei, KONGLingxuan, LIUKun, et al. Drone radio frequency signal identification method based on MobileNet-DOA[J]. Radar Science and Technology, 2025, 23(1): 57-66. (in Chinese)
TAOWenbo, NIDongming, CHENGYu, et al. Multi-feature fusion unmanned aerial vehicle sound target recognition method based on CNN-LSTM-PSA[J]. Audio Engineering, 2025, 49(7): 153-159. (in Chinese)
[7]
HUANGY, QUJ, WANGH, et al. An all-time detection algorithm for UAV images in urban low altitude[J]. Drones, 2024, 8(7): 332.
[8]
AL-EMADIS, AL-ALIA. Audio-based drone detection and identification using deep learning techniques with dataset enhancement through generative adversarial networks[J]. Sensors, 2021, 21(15): 4953.
[9]
AYDıNİ, KıZıLAYE. Development of a new light-weight convolutional neural network for acoustic-based amateur drone detection[J]. Applied Acoustics, 2022, 193: 108773.
XIAHao, LINYuewei, YANJun, et al. Research on key technologies of low altitude unmanned aerial vehicle for integrated sensing and communication: an overview[J]. Telecommunication Engineering, 2025, 65(6): 838-847. (in Chinese)
SUNYongsheng, LEIZonglin, LIEnte. Electronic data forensics technology for UAV involved in illegal and non-compliant flights[J]. Radio Engineering, 2025, 55(9): 1887-1893. (in Chinese)
QIAOQian, YINTong, ZHANGQiuyun, et al. Research on UAV classification techniques based on time-frequency features[J]. Journal of Ordnance Equipment Engineering, 2025, 46(7): 183-192. (in Chinese)
ZHAOQiyun, ZHANGHe, JINChong, et al. “Low-slow-small” UAV recognition and tracking algorithm based on improved model YOLO11-DeepSORT[J]. Fire Control & Command Control, 2025, 50(7): 125-132. (in Chinese)
WANGManlin, ZHUXiubin, YANGLan, et al. Multi-stage cross-view UAV image matching via hierarchical feature fusione[J/OL]. Journal of Frontiers of Computer Science and Technology, 2025: 1-19. (2025-07-11).in Chinese)
[20]
ALLAHHAMM S, AL-SA’DM F, AL-ALIA, et al. DroneRF dataset: a dataset of drones for RF-based detection, classification and identification[J]. Data in Brief, 2019, 26: 104313.
[21]
HEK, ZHANGX, RENS, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.
[22]
SZEGEDYC, IOFFES, VANHOUCKEV, et al. Inception-v4, inception-ResNet and the impact of residual connections on learning[C]//Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence, 2017: 4278-4284.
[23]
TANM, LEQ V. EfficientNet: rethinking model scaling for convolutional neural networks[J/OL]. Computer Science, 2019.
[24]
DUH, CHEND, ZHANGX, et al. A dynamic gesture recognition method based on MobileNetV4[C]//2024 8th Asian Conference on Artificial Intelligence Technology (ACAIT), 2024: 54-60.